A catalog needs to help customers find products and help the team classify the next product correctly. Start by assigning category, product type, vendor, and tags clear jobs. Then build curated and rule-driven collections and verify sorting, menus, search, filters, and the product data used by feeds.
9 parts · 3 checkpoints
Show all 9 parts
- 01Map the smallest useful customer path
- 02Give every product field a stable job
- 03Identify the collection model before choosing a workflow
- 04Test both matches and non-matches for all / any
- 05Complete the six catalog jobs in admin
- 06Separate membership, filtering, and customer entry
- 07Practise repairing recurring temporary-tag mistakes, has a checkpoint
- 08Keep the catalog, tests, and maintenance record, has a checkpoint
- 09Work it out: which catalog structure to fix first
Key point Establish why a product belongs in the collection
Correct membership now does not establish correct future classification or synchronized menus, feeds, and search platforms. Keep the catalog map, field dictionary, rules, positive and negative tests, and customer path. Tutorial checks record your work; they do not change the store or run tests.
A product feed is product data sent to a sales or advertising channel. Completing product fields in the store does not show whether the channel received the data or synchronized it.
The main paths are Products > Collections, Content > Menus, and Apps > Search & Discovery. Check the intended store, complete section, permissions, plan, and save response before proceeding. The series example plans a desk-organization store with one physical product, Variant A/B, a United States launch, USD, and an English storefront. These are exercise inputs, not evidence that products, channels, or earlier lessons are complete. Prepare at least one complete actual product and draft its fields first.
Questions this lesson answers
- When should you choose manual or automated collections?
- What do category, product type, vendor, and tags each do?
- How do all/any tests prevent unwanted inclusions and missing products?
- How do you check title, description, image, SEO, and sorting together?
- Why is a new collection missing from the main menu?
- How do Search & Discovery filters avoid empty results?
- How can the catalog grow while its fields remain stable?
- How should collections, menus, filters, and advertising feeds share field definitions?
This exercise needs four draft test products (two brands, two types), all unpublished and excluded from every sales channel, deleted afterwards; with only one product, create drafts first.
Map the smallest useful customer path
A collection groups related products on a page customers can visit. They open a product from that page and choose its variant on the product page. A classification field is separate from a menu, and a collection still needs a menu link. A product can belong to several collections when each has a clear purpose.
Sketch three layers: primary navigation category → collection page → product page. This is a way to understand the shopping path, not a Shopify limit on menu depth. Consider All Products, Desk Organization, and New Arrivals for browsing everything, shopping for desk organization, and discovering new items. With a small launch assortment, their products might overlap. Record their purpose, current products, and realistic future capacity before deciding which deserve menu space.
Desk Organization should introduce products that solve that shopping task. If New Arrivals is curated, specify who adds products and when they leave. Merge vague collections without products or a distinct task instead of filling the catalog with internal department names or a separate, indistinguishable collection for every product. Keep URLs, selection mode or source, owner, and verification time. Future additions still require tests.
Settle these four catalog decisions
Launch collection
- Decision and example
- Desk Organization serves a clear desk-organization shopping task.
- Reason and verification
- It has sellable products or a defined expansion reason, and the menu label matches the task.
Curation
- Decision and example
- Campaigns, selected products, or small assortments with unstable rules.
- Reason and verification
- A person controls content, addition, removal, sorting, and review.
Rule-driven collection
- Decision and example
- Stable category or product-type fields define the assortment.
- Reason and verification
- New products join by rule. Test matches, non-matches, and field changes before calling it complete.
Tags
- Decision and example
- Small, controlled operational labels with addition/removal responsibility.
- Reason and verification
- They do not replace standard category or stable product type, or contain whole product descriptions.
Enter actual store values row by row. Mark a decision passed only with saved or tested evidence; return to the relevant step for missing permission, ownership, or results. A store with one product can start small. An admin count cannot establish useful navigation or marketing performance.

Historical Products → Collections example. Automated Collection, Hydrogen, and Home page are development data. The price-range example is separate from the classroom Desk Organization plan. Counts and condition summaries do not establish correct membership, storefront access, or feed sync.
Give every product field a stable job
Open Products > current product and inspect Category and Product organization. Category comes from Shopify's Standard Product Taxonomy and describes the product. Product type is the team's custom type. Vendor identifies its brand, manufacturer, or supplier. Tags are searchable keywords that can also drive collection rules and operations. They are not exclusively internal: search, themes, or filters can use them, and customers may see relevant values.
| Field | Classroom input | Owner and use |
|---|---|---|
| Product category | Choose the most accurate standard category by the product's main purpose; do not substitute a made-up collection name | Product-data owner; standard attributes, conditions, and channel taxonomy mappings need their own checks |
| Product type | Use Desk Organizer consistently instead of competing spellings for the same type | Catalog owner; stable rules and applicable filters |
| Vendor | Example Brand is a placeholder for the actual brand or supplier | Product-data owner; brand filters and channel data |
| Tags | launch-review identifies an owned operational state with a review expiry | Operations owner; document which tools read it and who adds/removes it |
These are proposed values, not saved store records. Copying Desk Organization into all four fields does not define their responsibilities. Record definitions, allowed values, naming conventions, consumers, and owners, then verify actual values on at least one product. Collections, store search, filters, and advertising product groups should use consistent definitions. Each feed's mapping, receipt, and synchronization still needs a separate channel check; tidy fields do not establish accepted data.
Identify the collection model before choosing a workflow
Shopify is rolling out a new collection model. Stores without it can still use legacy Manual and Smart collections, called Automated in older material. Inspect the complete current screen before looking for historical buttons. In the new interface, a Products source offers Add products, Add condition, and applicable Exclude controls. The legacy creation flow selects Manual or Smart. Both workflows need an explanation of why those products belong together.
Manual selection suits campaigns, small curated groups, and rules that cannot be expressed reliably. Use owned automatic conditions for stable categories that will grow. In the new model, a Products source includes the whole product and its variants. To include only Variant A rather than Variant B, use an available Variants source and test the result. Legacy variant conditions do not establish the same variant-level presentation. Record model, source, product/variant scope, and available controls; investigate missing functionality before assuming support.
Test both matches and non-matches for all / any
Match all conditions acts as AND: every condition must hold. Match any condition acts as OR: one match is sufficient. For one classroom source, propose Type equals Desk Organizer and Vendor equals Example Brand. The table uses proposed test products; it does not establish that they exist in the store.
| Test product | Type matches | Vendor matches | Included with all | Included with any |
|---|---|---|---|---|
| Example Brand desk organizer | Yes | Yes | Yes | Yes |
| Future product of another type from the same brand | No | Yes | No | Yes |
| Desk organizer from another brand | Yes | No | No | Yes |
| Another type from another brand | No | No | No | No |
For “this brand's desk organizers,” any also admits the middle two groups. Write expected results in the product sheet before saving, then compare the actual membership and count. Category, type, vendor, tag, price, and inventory can support applicable rules, but check the current field, operator, and value. An upper and lower price or inventory range needs all, because any can admit values outside the intended range. The historical screenshot's price bounds illustrate this logic; they are not a pricing recommendation for this store.
In the new model, inspect every source and exclusion. Another source or a manual addition can explain extra products; exclusions can override manual or automatic inclusions. To test a controlled catalog-hold tag exclusion, first establish that the product qualifies for inclusion, confirm that the exclusion removes it, then reverse the test change and confirm the expected result returns. Assign the tag an owner and purpose; it does not automatically mean unavailable or out of stock. Use the controls your legacy interface actually provides rather than assuming a manual deletion will permanently remove an automatically matching product.
Keep added, missing, unwanted, and excluded-product results. Under a safe test arrangement, change a product field, observe membership, and restore the test value. Do not broaden tag scope without an owner or hide missing inputs with a one-off manual addition. A correct test set establishes only that these inputs and rules agree now.
Complete the six catalog jobs in admin
The person authorized to maintain the catalog should act in the intended store. Save and reopen each change, then verify the storefront. Retain actual values, time, owner, and evidence location. Use these steps for both curation and automation, choosing the entry for the current model.
Map the three-layer catalog
Where in the UI: ProductsCollections
Connect primary navigation, collection pages, and product pages for candidates such as All Products, Desk Organization, and New Arrivals. State each shopping intent, current assortment, and realistic expansion capacity.
You should see
Have a minimum catalog sketch, collection URLs or pending-creation labels, owners, and review time.How to confirm
Every entry has a customer task and products or a specific expansion reason. A candidate entry is not automatically a published menu item.If it goes wrong
Merge vague or excessive collections around shopping tasks. Remove purposeless entry plans rather than adding internal department names to fill space.Keep as evidence
Keep the map, current product coverage, URLs, owners, and time; future products still need verification.Separate field responsibilities and verify product values
Where in the UI: Productscurrent product
Use the dictionary for standard category, stable product type, actual vendor, and controlled tags. Save and reopen. Normalize names and record which collections, filters, and feeds consume each field.
You should see
Each field has a definition, allowed values, owner, and use; at least one complete product has reviewable actual values.How to confirm
Check membership, store search, and applicable filters with actual products, and inspect channel mappings separately.If it goes wrong
For wrong membership, test each condition, all/any, and field value. If long-term rules all depend on temporary tags, assign maintenance and use stable fields.Keep as evidence
Keep the dictionary version, actual-value summary, match result, and owner. Exclude sensitive customer or supplier data from public evidence.Create a curated collection
Where in the UI: ProductsCollectionsAdd collection or Create collection
Enter a title, unique description, image, and search listing. In the new model, select items through
Add productsin a Products source; in legacy, select Manual and add the products. Save and inspect sorting, URL, sales-channel availability, and the first storefront screen.
Historical English development interface: the collection is unsaved and Collection items is zero. This shows source and selection controls, not saved conditions or matching results. It is not the old Manual/Smart choice form.
You should see
Have a curated collection with a theme, copy, image, product list, and sorting result.How to confirm
Selected products, URL, and first-screen content match the shopping task, and customers can open the intended page.If it goes wrong
For an empty storefront page, check product status, product/collection channel publishing, market availability, sources, and exclusions. Do not repeatedly add the same product.Keep as evidence
Record model, manually selected scope, product count, sorting, image, URL, and storefront test.Create and test automatic conditions
Where in the UI: ProductsCollectionscurrent collection
Use
Add conditionin the correct source of the new model, or choose Smart/Automated in the legacy flow. Inspect field, operator, value, and all/any before running the positive and negative tests. Check Variants and Exclude where applicable and whether other sources widen the result.You should see
Have explicit, currently correct conditions, a stable-field maintainer, and a way to review future additions.How to confirm
Membership changes as expected after adding or editing a test product. Retain inclusion, non-inclusion, exclusion, and restoration results.If it goes wrong
For missing products, check populated fields, conditions, source, exclusion, and update state; for an empty storefront, also check publishing and market. Recheck cache after testing, without broadening tags or hiding broken rules through manual additions.Keep as evidence
Keep rule, model/source, test products and expected/actual results, match count, URL, maintainer, and review time.Set sorting, SEO, and mobile presentation
Where in the UI: Current collectionsorting and Search engine listing
A new store can curate manually to put suitable, available products first. Best selling is an option once sales inform the order, but it does not guarantee that unavailable products move down. Current guidance says that with no sales it sorts newest to oldest, so inspect the actual sequence. Write collection-specific copy explaining who it serves, how to choose, and how it differs from other groups. Check image, search-listing preview, and URL instead of copying a product description.

The historical form shows image/description entry, Theme template, and Search engine listing. It does not show saved sorting or mobile results. Verify independent copy, image, SEO, and the first screen in your own collection.
You should see
Have suitable sorting, independent content, image, SEO, and an actual mobile display result.How to confirm
Check theme presentation, available first-screen products, image crop, and filters on the mobile storefront. A search preview does not establish indexing or ranking.If it goes wrong
Adjust an unsuitable order and retest. For excessive or empty filters, normalize product values, retain customer-useful fields with complete coverage, and remove or hide low-value filters.Keep as evidence
Keep sorting, title/description/SEO summary, image, URL, mobile captures, and filter results.Connect menus, filters, and feeds, then follow the customer path
Where in the UI: ContentMenus; AppsSearch & Discovery
Confirm online-store availability, add the collection link to the intended menu in Menus, save, and establish that the theme uses that menu. In Search & Discovery, use
Filters > Add filterto select and save needed sources; the theme must support filtering. Inspect feed or advertising product-group mappings and channel receipt rather than maintaining a second field system with conflicting meanings.You should see
The menu reaches the collection, filters work, product pages open, and feed fields agree with the dictionary.How to confirm
On desktop and mobile in the target market/language, use menus or store search, apply and clear filters, and open the correct product and variant. Read channel synchronization back separately.If it goes wrong
For a missing menu entry, inspect saved link, collection publishing, and theme menu placement. For missing filters, inspect theme support, source, and product values. For empty combinations, check field coverage and selection logic before introducing another field system.Keep as evidence
Record menu position, collection link, search term, filter combinations, product/variant, market/language, feed mapping and receipt state, time, and owner.
Separate membership, filtering, and customer entry
Collection conditions determine membership. Filters refine the current collection or search results. Different filters normally combine with AND; multiple values within one filter normally default to OR, with configurable behavior for some sources. These controls are separate from a collection source's all/any setting. Test a single filter, combinations, and restoration after clearing them using actual product data. Missing values, inconsistent spelling, or poor coverage can hide useful products. A configured filter also needs a compatible theme to appear.
| Symptom | Return to | Verify before continuing |
|---|---|---|
| Unwanted or missing collection products | Dictionary, all/any, all sources, manual additions, and exclusions | Compare positive/negative cases, change a test input, restore it, and keep expected/actual results |
| Admin has products but the storefront collection is empty | Status, product/collection channels, market, and rules | Open the URL in the target market and then reach it through the menu; do not mask the cause with repeated manual additions |
| Collection has no menu entry | Saved menu, link target, collection availability, and theme menu | Follow menu → collection → product on the storefront, beyond an admin preview |
| Too many filters or empty results | Theme support, data coverage, normalized values, and combined logic | Retain useful, complete filters; test single values, combinations, and clearing. Use current help for unresolved indexing updates |
| Expansion breaks classification | Required stable fields, naming, and maintenance ownership | Repair inputs and responsibility, then retest existing and new products instead of only deleting mistakes |
Record product discovery, collection classification, feed receipt, and external indexing separately. Collection SEO and internal links have distinct jobs, but saving does not guarantee indexing or conversion. Before edits, retain original conditions, manual lists, exclusions, sort, and links. Restore the affected change if needed and retest the customer path. Avoid simultaneous edits to the same collection.
Practise repairing recurring temporary-tag mistakes
Choose actual products and test a stable product type or another owned field. Temporary campaign tags need a person who adds and removes them. In the scenario below, a new product is already misclassified. Whether a wanted product is missing or an irrelevant one appears, inspect the rule, input, and maintenance responsibility first.
Choose a maintainable repair
Self-check not answered
Also inspect collection SEO, sorting, and mobile presentation, and record menu, search, filter, and feed results and follow-up work. These controls do not save Shopify settings or perform tests.
Actual conditions for expanding the catalog
| Done when | Stop if |
|---|---|
Collections serve real shopping tasks, field jobs are documented, conditions are explainable and tested, sorting/SEO/mobile/menu/search/filter results are checked, and channel mappings and synchronization are recorded separately. | Rules rely on ad-hoc tags, fields have conflicting meanings, products are misclassified, storefront pages are empty, or fields and rules have no maintainer. |
| Evidence Minimum map, dictionary, actual values, curated/rule lists, expected/actual tests, collection URLs, customer paths, feed records, and accountable owners. | |
Keep the catalog, tests, and maintenance record
Keep a minimum usable catalog, including applicable All Products, Desk Organization, and New Arrivals purposes, URLs, models/sources, conditions, curated lists, exclusions, sorting, images, and storefront entries. Retain the field dictionary, test-product expected/actual results, mobile/menu/filter observations, feed mapping and receipt, blockers, verification time, and maintainer. Mark unverified work pending.
The record saves automatically in this browser, reusing the decision and checks above. Enter redacted summaries and locations of controlled documents, keep longer product sheets separately, and export or copy before changing devices. Resetting the lesson record does not undo Shopify product fields, collections, or menus.
What this record establishes
Hold
Choose a repair first
Repair rules and ownership
Some checks remain
Decision and checks recorded
Once rules and customer paths are clear, duplicate the theme and establish customization, preview, publish, and rollback procedures. Recheck collection and product entry after theme changes.
No checkpoint finished in this lesson yet. Start with “Go to the exercise”.
Self-check not answered
FAQ
- Does a larger number of collections look more professional?
Start with a minimum catalog customers understand, with products now or a defined expansion reason. Empty collections and single-product groups without distinct purposes add navigation and maintenance work.
- Can automated collections rely only on tags?
Controlled tags can support a temporary campaign with addition, removal, and review owners. Prefer stable category or product type for long-term categories. Tags may also support customer search or filtering, so avoid exposing sensitive internal meanings.
- Why is a new collection missing from navigation?
Collections and menus are separate settings. Save a collection link in Content > Menus, establish online-store availability and the theme's selected menu, then test the storefront.
- When should I use automated versus manual collections?
Use conditions for stable rules and growing assortments; select items manually for small curated groups, campaigns, or unreliable rules. Follow the current source controls or legacy Manual/Smart interface rather than treating historical buttons as universal.
- Can tags be the main product taxonomy?
Tags serve keywords, controlled rules, and operations. Standard category and stable product type provide long-term classification so filters and feeds use consistent definitions. Repeating every value across all fields does not design those responsibilities.
- Does a collection page need its own description?
Explain who the collection serves, how to choose, and how it differs from others, with independent SEO and internal-link content. Avoid copying product descriptions. Previewing and saving do not establish indexing or ranking.
Continue with the problem you have now
Work it out: which catalog structure to fix first
Four catalog structures (example data). Rule: a collection is worth what it lets someone find; empty collections and a single "everything" collection both fail that test, and differ only in what it costs to repair.
Collections and products in four structures
| Item | Collections | With products | Problem |
|---|---|---|---|
| Structure A | 30 | 10 | 20 are empty and still in the nav |
| Structure B | 8 | 8 | None |
| Structure C | 12 | 12 | Three groups overlap, so products repeat |
| Structure D | 1 | 1 | One "all products" collection carries every SKU |
Decide before you read on. Say which one to fix first, and why D, with the fewest collections, is not the simplest of the four.
Answer key: structure A comes first — twenty empty collections still sit in the navigation, sending visitors into "no products" pages again and again, and that loss happens immediately. C repeats the same products across overlapping sets, which distorts filtering and sorting while people can still find items. D has one collection, so nobody can narrow by use: fewer is not better here, it means no forks in the road at all. D is the trap, because it looks the cleanest while moving the whole cost of choosing onto the visitor.